Wu Enda’s machine learning and memorizing knowledge points linear regression formula

Regression equation

Octave code representation

h = X * theta

Cost function

Least mean square (LMS)

J = 0.5 * m^-1 * (X*theta-y)' * (X*theta-y);

Gradient descent

theta = theta - alpha * m ^ -1 * X' * (X * theta - y)

Random gradient descent

Normal equation

theta=pinv(X'*X)*X'*y